🐝 Intel Swarm

Autonomous AI intelligence research across 19 domains — War, Crypto, Macro, AI Agents, Communist States, Health, Religion, Commodities & more. Signal over noise.

Live: intel-swarm.vercel.app · 繁體中文

---

What is this?

Intel Swarm is an autonomous intelligence research team. A swarm of specialized AI researcher agents scrape, analyze, and synthesize the web daily — delivering high-signal briefings across 19 domains in English and Traditional Chinese.

Each morning:

1. 19 researcher agents run in parallel across domains (06:00–07:45 HKT)
2. Synthesis agent connects the dots across all findings
3. Chief analyst delivers the final actionable briefing
4. Everything is translated to Traditional Chinese automatically
5. Published to the web at 08:00 HKT via automated git push + Vercel deploy
6. Breaking intel is continuously monitored and pushed to Telegram every 2 hours

---

Domains

Domain	Focus
⚔️ War	Active conflicts, geopolitical flashpoints, military movements
🛢️ Commodities	Gold, oil, copper, rare earths, supply chains, chokepoints
🔴 Communist States	Russia 🇷🇺 · China 🇨🇳 · North Korea 🇰🇵
✝️ Religion	Religious nationalism, Vatican, faith & geopolitics
🧬 Health	Biotech, longevity, pandemics, AI diagnostics
🎭 Culture	Gen Z, virality, social movements
🌍 Emerging	Southeast Asia, Africa, LatAm opportunities
🤖 AI Agents	Autonomous AI, frameworks, agentic products
🪙 Crypto	DeFi, Solana, on-chain signals
📊 Macro	Fed, BRICS, debt, currency wars
🧠 Singularity	AGI timelines, compute wars
📈 Quant	Market signals, volatility
🌏 West-East	China/HK signals the West misses
🖤 Black Budget	DARPA, CIA, UAP disclosure
🕳️ Conspiracy	Mainstream-bound narratives
📁 Epstein	Files, court docs, elite networks
🏆 Sports	Business side of sports

---

Features

🗞️ Daily Intelligence Feed

• Hero banner with top signal of the day
• Card grid with OG images, scores, and domain tags
• Grid / Feed toggle view
• Full bilingual support — EN + Traditional Chinese (繁體中文)
• Date selector to browse historical intel

📊 Polymarket Overlay

• Hover any finding to see live Polymarket prediction market odds
• Corner overlay with real-time probability chart
• Domain-aware matching — only causal, relevant markets shown
• Click to pin; "View on Polymarket ↗" direct link
• Toggle on/off via the P button in the header

🌍 Conflict Heat Map (/conflict)

• Interactive Leaflet map with intel findings as live pins
• GDELT event data as background heat layer
• Filter by domain: War · Politics · Intel · Culture
• Fully bilingual (EN/ZH labels, finding titles, domain names)

⚡ Bloomberg Terminal (/terminal)

• Bloomberg-style dark feed of all findings across all domains
• Domain filter panel (left) + findings feed (right)
• Ticker tags: $BTC $ETH $SOL OIL GOLD DXY BONDS SPX NVDA TSLA DEFENSE + more
• Signal strength pips, real-time search
• Built-in API docs modal

🔌 Public API v1

Free, no auth required. Base URL: https://intel-swarm.vercel.app

Endpoint	Description
GET /api/v1/domains	List all research domains
GET /api/v1/findings	All findings (filter by date, domain, lang, limit, ticker)
GET /api/v1/feed	Score-sorted feed across all domains
GET /api/v1/brief	Latest chief analyst brief
GET /api/v1/search?q=	Full-text search across findings

📡 RSS Feeds

• GET /rss — Global feed (all domains, 50 items)
• GET /rss/<domain> — Domain-specific feed

📲 Breaking Intel Push

• Scans for new findings every 2 hours
• Formats Telegram messages with ticker tags and source links
• Requires TG_PUSH_BOT_TOKEN + TG_PUSH_CHANNEL_ID env vars

🌙 Light / Dark Mode

• Toggle via ☀️/🌙 button in the header
• Persists across sessions via localStorage

---

Stack

• Agents: OpenClaw — AI agent platform running cron jobs
• Models: Claude Opus 4.6 (synthesis & chief), Claude Sonnet 4.6 (domain research), Claude Haiku (translation)
• Web: Python / Flask, deployed on Vercel
• Map: Leaflet.js + GDELT 2.0 CSV feed
• Prediction Markets: Polymarket Gamma API + CLOB API
• Images: OG image scraping from source URLs
• Languages: English + Traditional Chinese (繁體中文)

---

Setup

Prerequisites

• OpenClaw installed and configured
• Vercel account + CLI (npm i -g vercel)
• Claude API access via OpenClaw

1. Clone & install

git clone https://github.com/outsmartchad/intel-swarm
cd intel-swarm
pip install -r requirements.txt

2. Configure environment

cp .env.example .env
# Edit .env — see .env.example for all required variables

3. Create all researcher crons

export TELEGRAM_USER_ID=your_telegram_id
python3 create-crons.py

4. Run locally

cd web && python3 server.py
# Open http://localhost:5757

5. Deploy to Vercel

vercel deploy --prod --yes

⚠️ This project uses a legacy builds config in vercel.json. GitHub pushes alone do not trigger Vercel redeploys — always use vercel deploy --prod --yes.

---

Daily Pipeline

06:00 HKT  →  19 researchers run in parallel across domains
07:00 HKT  →  Synthesis + Chief briefing generated
07:45 HKT  →  Chief delivers Telegram briefing
08:00 HKT  →  Autopush: copy → translate → scrape images → git push → vercel deploy
Every 2h   →  Breaking intel scan → Telegram push

---

Project Structure

intel-swarm/
├── researchers/            # Per-domain researcher agents
│   └── {domain}/
│       ├── findings/       # Daily MD files (EN + ZH)
│       └── memory/         # Persistent threads & sources
├── synthesis/              # Synthesis agent
├── chief/                  # Chief analyst agent
├── web/
│   ├── server.py           # Flask app — all routes, API, RSS, Terminal, Conflict
│   ├── templates/
│   │   ├── base.html       # Masthead, nav, theme, Polymarket overlay
│   │   ├── index.html      # Homepage — hero, grid, feed toggle
│   │   ├── domain.html     # Domain page with sub-tabs + findings
│   │   ├── conflict.html   # Leaflet conflict heat map
│   │   └── terminal.html   # Bloomberg-style terminal
│   ├── static/
│   │   ├── conflict-events.json   # GDELT heat layer data
│   │   └── push-state.json        # Breaking intel push state
│   ├── translate.py         # EN → Traditional Chinese translation
│   ├── scrape-images.py     # OG image scraper
│   ├── prefetch-polymarket.py    # Polymarket market prefetch pipeline
│   ├── prefetch-conflict.py      # GDELT CSV fetch + parse pipeline
│   ├── push-breaking.py          # Breaking intel Telegram push
│   ├── translate-all.sh          # Batch translate all domains
│   └── scrape-all-images.sh      # Batch image scrape all domains
├── docs/
│   ├── worldmonitor-api-intelligence.md     # API source catalogue (EN)
│   └── worldmonitor-api-intelligence.zh.md  # API source catalogue (ZH)
├── create-crons.py          # Spin up all OpenClaw crons
├── .env.example             # Config template
├── CONTRIBUTING.md
└── CONTRIBUTING.zh.md

---

API Reference

Environment Variables

Variable	Required	Description
BRAVE_API_KEY	✅	Brave Search API key for researchers
TELEGRAM_USER_ID	✅	Your Telegram user ID for briefings
TG_PUSH_BOT_TOKEN	Optional	Bot token for breaking intel channel push
TG_PUSH_CHANNEL_ID	Optional	Channel ID for breaking intel push

See .env.example for full list.

---

Contributing

Intel Swarm is designed to be extended. You can:

• Add new research domains — new topic, new researcher, automatic pipeline integration
• Refine researcher prompts — tune search queries, source preferences, output format
• Modify synthesis/chief logic — change how findings are connected and delivered
• Add sub-domain tabs — group related topics (e.g. Russia / China / North Korea under Communist States)
• Integrate new data APIs — see docs/worldmonitor-api-intelligence.md for 31 catalogued free APIs

Full guide: CONTRIBUTING.md · 中文說明 CONTRIBUTING.zh.md

---

License

MIT — fork it, build on it, run your own swarm.

---

Built with OpenClaw · Last updated: 2026-03-05